The amount of data created by pharmaceutical companies has increased exponentially over the last few decades. Managing these enormous volumes of data can be challenging, an issue compounded by a lack of standardisation. Data is often gathered from numerous sources and devices and stored separately, making it difficult to retrieve, query, share and analyse.
FAIR data is data that meet principles of findability, accessibility, interoperability, and reusability. FAIR Data Principles were created to increase standardisation and harmonisation across the industry and allow more efficient management, processing and analysis. Supporters argue that FAIR DATA has the power to transform the industry and accelerate faster drug discovery and development, but the pharma industry has been slow to adopt it. Implementing FAIR Principles can be a daunting task; it is rarely a quick and straightforward process, as it requires a shift in how data is created and used within an organisation.
Where Are We Now?
There are significant technical and cultural barriers to FAIR data implementation for pharmaceutical companies. Despite this, the benefits far outweigh the problems for most organisations and make data optimisation worthwhile. Three of the biggest obstacles to FAIR data implementation are initial costs, attitudes towards data management and resistance to changing workflows.
Misinterpretations: What is FAIR Data?
As FAIR data is a relatively new term, it has been the subject of a lot of discussions. To understand its concept, we should examine some misconceptions associated with it. FAIR is not a standard but instead a set of principles. Standards are prescriptive, while principles allow room for interpretation and customisation. For data to be FAIR, it merely needs to follow the fundamental principles of being Findable, Accessible, Interoperable, and Reusable. Companies can and should create standards that make sense for their data assets.
FAIR is also not solely based on making data more manageable for humans. Instead, one of the critical drivers for FAIRification is to increase how effectively AI and ML applications can analyse data.
Finally, and perhaps most importantly, FAIR data does not have to be open access. There are many reasons that data may need to be kept private, including personal details, national security, and competitiveness. The ‘A’ in FAIR stands for “Accessible under well-defined conditions”, and these conditions can vary depending on the source and use cases of data.
Examining the Cost: Implementation and Returns
FAIRification can be a costly task, requiring updated infrastructure time, effort, and training. Applying FAIR data principles to retrospective data can be a particularly daunting task. Despite this, research from the European Commission has estimated the minimum annual cost of not having FAIR data as €10.2bn across the European Union despite the initial costs. In addition, the benefits of improved data management have long-term financial benefits. Aligning data with FAIR principles can positively impact pharmaceutical organisations by increasing the value and insights gained from their data assets. Additionally, FAIRificartion helps improve the effectiveness of AI and machine learning. In turn, this can lead to benefits across the entire value chain, from discovery and development to patient outcomes.
Adjusting Attitudes: Data Management and Workflows
While FAIRification of data is generally agreed to be beneficial, it has met resistance from some scientists. Updating data is often viewed as an additional task that reduces productivity and takes away the time scientists can spend on furthering scientific research. Alharbi et al. (2021) surveyed scientists to improve their understanding of attitudes towards FAIR data. They found that “respondents asserted that the time spent aligning data with FAIR principles might affect an individual’s productivity and thereby significantly influence a company’s day-to-day business of drug discovery. They stated that the FAIRness of the data is not their top priority, but rather a priority secondary to the scientific progression of the project.”
Attitudes towards FAIR data are only likely to change when the benefits become more apparent. The acronym and principles were very recently defined in a March 2016 paper in the journal Scientific Data. It will take some time before insights uncovered through the FAIRification of data become more widespread. Additionally, once workflows have shifted to a FAIR Data pipeline, most research scientists will find that they spend less time managing data.
The pharmaceutical industry is becoming ever more data driven as regulatory requirements increase and trial data becomes more complex. FAIR data principles provide guidelines on organising and managing data and increasing the value and insights gained from data assets. The implementation of FAIR also has the potential to enable greater collaboration and streamlined data management, opening doors to new innovations and insights.
Alharbi, E., Skeva, R., Juty, N., Jay, C. and Goble, C., 2021. Exploring the Current Practices, Costs and Benefits of FAIR Implementation in Pharmaceutical Research and Development: A Qualitative Interview Study. Data Intelligence, 3(4), pp.507-527.
Mons, B., Neylon, C., Velterop, J., Dumontier, M., da Silva Santos, L. and Wilkinson, M., 2017. Cloudy, increasingly FAIR; revisiting the FAIR Data guiding principles for the European Open Science Cloud. Information Services & Use, 37(1), pp.49-56.